Combinatorial Genomics in Cancer

描述(申请人提供)：癌症的靶向治疗需要对驱动恶性细胞生长的基因改变有清楚的了解。癌症病因学的三个特点使病因基因改变的鉴定变得复杂：1。通常需要多种相互作用的改变才能导致癌症。几个不同的变化可能足以产生一个单一的癌症表型，和3。致癌改变出现在正常遗传活动和恶性细胞生长的虚假后果的密集背景中。我们建议将机器学习算法Prim的一种变体应用于识别导致特定癌症或提供有关临床病程和治疗效果的预后信息的分离的连接性基因改变集的任务。这些分析从我们已经开发的低水平生物信息学资源中合成信息，以绘制染色体变化图并监测转录因子活性的全球模式。基于这些基础，本研究开发了高水平的分析工具来绘制低水平基因组事件之间的组合相互作用图。具体地说，这些研究试图：目标1：开发图形用户界面(GUI)软件，以支持计算背景有限的生物学家进行组合基因组分析。目的2：优化疾病进展和治疗反应的组合预测。目的3：建立基于Prim的统计模型，以确定基因改变和转录控制信号的功能互补组。这些研究中产生的生物信息学工具将创建一个通用的分析基础设施，用于绘制癌症的复杂病因图，并部署针对患者的靶向治疗。

DESCRIPTION (provided by applicant): Targeted therapy of cancer requires a clear understanding of the genetic alterations that drive malignant cell growth. Identification of causal genetic alterations is complicated by three characteristics of cancer etiology: 1.) multiple interacting alterations are often required to cause cancer, 2.) several distinct alterations may be sufficient to generate a single cancer phenotype, and 3.) oncogenic alterations appear in a dense background of normal genetic activity and spurious consequences of malignant cell growth. We propose to apply a variant of the machine learning algorithm PRIM to the task of identifying disjunctive sets of conjunctive genetic alterations that cause specific cancers or provide prognostic information about clinical course and treatment efficacy. These analyses synthesize information from low-level bioinformatics resources we have already developed to map chromosomal alterations and monitor global patterns of transcription factor activity. Based on those foundations, the present studies develop high-level analytic tools to map combinatorial interactions among low-level genomic events. Specifically, these studies seek to: Aim 1: Develop graphical user interface (GUI) software to support combinatorial genomic analyses by biologists with limited computational background. Aim 2: Optimize combinatorial prediction of disease progression and treatment response. Aim 3: Develop PRIM-based statistical models to identify functional complementation groups of genetic alterations and transcriptional control signals. The bioinformatic tools produced in these studies will create a generalized analytic infrastructure for mapping complex etiologies in cancer and deploying patient-specific targeted therapies.